Principal AI & Salesforce Architect - Remote in US (Remote, TX, US)

About this role

Req ID:393533
NTT DATA strives to hire exceptional, innovative and passionate individuals who want to grow with us. If you want to be part of an inclusive, adaptable, and forward-thinking organization, apply now.
We are currently seeking a Principal AI & Salesforce Architect - Remote in US to join our team in Remote, Texas (US-TX), United States (US).
Principal AI & Salesforce Architect
Forward Deployed Engineering · Senior, client-facing · Full time · Regular travel to client sites
The opportunity
Agents are becoming the users of enterprise software. Salesforce is responding by moving its value beneath the screen: capabilities any agent can invoke, business logic packaged as Skills, trusted context in Data 360, orchestration across vendors in Agent Fabric, and a choice of models from Salesforce's own CRM reasoning model to the frontier labs.
Most enterprises can see the shift. Very few have someone who can explain where AI is going, decide what it means for their business and their Salesforce estate, and then make it work in production. This is a strategic, AI-forward role that stays with the client while uncertainty and risk are highest, remains through production, and hands ownership to the delivery team once the solution is proven and stable.
Not a typical FDE role
Most Forward Deployed Engineers embed one product in one company. Ours is a strategic seat in a global Salesforce practice.

•
Wider. You work across industries and our most strategic clients, not inside a single deployment.

•
Earlier. You shape the question with client executives before anyone designs a solution.

•
Further ahead. You lead the conversation about where models, agents and Salesforce are heading, not only what ships this quarter.

•
Hands-on. You prototype, debug, inspect code, understand failure modes and work directly with engineers to unblock the hardest problems.

•
Lasting. What you learn in the field becomes how the whole practice sells and delivers.

What you will do

•
Be the AI voice in the room. Brief client executives on where frontier and open-weight models, agents and Salesforce are heading, and translate that direction into business and technology roadmaps.

•
Shape our most strategic pursuits. Frame the problem, challenge whether an agent is the right answer, set the architecture and pressure-test scope, estimates, operating assumptions and risk.

•
Choose the right intelligence for each step. Decide where a frontier model, an open-weight model, a CRM reasoning model, deterministic logic or a person belongs. Weigh accuracy, cost, latency, privacy, sovereignty, hosting, operability and failure modes – not novelty.

•
Design agentic systems. Architect orchestration, tool use, retrieval and grounding, memory, evaluation and guardrails across Agentforce, Data 360 and external model platforms, and define how agents are tested, monitored and improved once they are live.

•
Design trusted agent authority. Define agent identity, delegated authority, tool access, action boundaries, approval checkpoints, auditability and revocation so agents act with least privilege, mapped onto the Salesforce security model and the client’s enterprise identity.

•
Architect across the estate. Translate business, functional and non-functional requirements into Salesforce designs, data models and integration patterns spanning Agentforce, Data 360, MuleSoft, Databricks and the wider enterprise estate, protecting platform integrity and maintainability.

•
Make it real, fast. Prototype with AI-assisted tools and our engineers; inspect and debug code, traces and integrations as needed so clients see working behavior and failure modes early, not a promise.

•
Lead delivery while the risk is highest. Set direction with architects, unblock the hardest problems and lead escalations across Salesforce, MuleSoft, Data 360, models and enterprise systems, alongside delivery teams and Salesforce engineering.

•
Stay through production. Remain engaged through production validation and stabilization, using real operating behavior to test the architecture and help restore stability when the hardest issues surface.

•
Hand off, then move on. Once the solution is proven and stable, transition architecture decisions and the reasoning behind them, known risks, runbooks and ownership to the delivery or managed-service technical lead and take on the next frontier problem.

•
Lead our thought leadership. Publish, speak and turn field patterns into the reference architectures, decision frameworks and approaches the whole practice uses.

•
Raise the bar. Mentor architects and engineers on AI and agentic design so the practice’s depth grows with every engagement.

What this role is not

•
Not a demo circuit. You stay with clients into production, where credibility is earned.

•
Not a coding seat, and not a hands-off architect. You prototype, debug, inspect code and unblock; our engineers build and own sustained development.

•
Not the delivery technical lead. You do not direct day-to-day engineering execution, run the code-review queue or own CI/CD and release management. You lead while risk is highest and then hand off.

•
Not delivery management or sales. Engagement managers run delivery and sales owns the commercial outcome; you own the technical truth.

•
Not a permanent fixture on one account. You go where uncertainty is highest, remain through production proof and stabilization, then transition ownership.

What you will bring

•
Real AI depth. Frontier models such as Claude, GPT and Gemini, and open-weight models such as Llama, Nemotron and Mistral: their trade-offs in capability, accuracy, cost, latency, privacy, sovereignty, hosting and failure behavior, and when each belongs in an enterprise.

•
Command of agentic design patterns. Tool use, planning, multi-agent orchestration, retrieval and grounding, memory, evaluation, guardrails and human oversight – including where each pattern fails in production.

•
A point of view on where AI is…